Paleoeskimo Demography on Western Victoria Island, Arctic Canada: Implications for Social Organization and Longhouse Development James M

Home , Boothia Peninsula, Lady Franklin Point, Victoria Island (Canada)

ARCTIC VOL. 65, NO. 2 (JUNE 2012) P. 167 – 181

Paleoeskimo Demography on Western Victoria Island, Arctic Canada: Implications for Social Organization and Longhouse Development James M. Savelle,1,2 Arthur S. Dyke,2,3 Peter J. Whitridge4 and Melanie Poupart1

(Received 15 April 2011; accepted in revised form 22 September 2011)

ABSTRACT. Paleoeskimo populations on western Victoria Island reached maximum levels in Early Pre-Dorset time and declined abruptly shortly after 3800 14C years BP. The largest subsequent recovery occurred during Dorset time, particularly during Late Dorset, about 1500 to 600 BP. Early Pre-Dorset settlement patterns were similar to those documented for the same period and culture elsewhere in Arctic Canada, with dispersed nuclear families and small extended families occupying the region for most of the year, but with annual aggregations producing sites with 15 or more dwellings. After 3800 BP, large Pre-Dorset aggregation sites are absent. Dorset settlement patterns are dominated by multi-family longhouse – hearth row aggregation sites and by sites with two to four dwellings. Possibly the Dorset were living mainly in snow dwellings on the sea ice during cold seasons, with longhouses and hearth-row sets representing coming-ashore aggregations. Architectural aspects of longhouses and hearth rows indicate a common purpose behind their construction and use throughout the region and apparently throughout Arctic Canada, but their place and time of origin remain obscure. Radiocarbon dates place most longhouses and hearth rows on western Victoria Island in Late Dorset time, as elsewhere, but some dates indicate that these structures were being used in the western Canadian Arctic by Middle Dorset time. The latest Dorset radiocarbon dates from the region overlap with the earliest of the more reliable dates on Thule houses. Key words: Paleoeskimo, Pre-Dorset, Dorset, Thule, demography, settlement patterns, longhouse, hearth row

RÉSUMÉ. Les populations paléoesquimaudes de l’ouest de l’île Victoria ont atteint leur summum à l’époque pré-Dorset puis ont chuté rapidement peu après 3800 années radiocarbones BP. Par la suite, la plus grande reprise s’est produite à l’époque du Dorset, plus particulièrement vers la fin de celui-ci, soit entre 1500 et 600 BP. Les types de peuplements du début du Dorset étaient semblables à ceux documentés pour la même période et pour la même culture ailleurs dans l’Arctique canadien, c’est- à-dire qu’il y avait des familles nucléaires dispersées et des petites familles étendues qui occupaient la région pendant la plus grande partie de l’année, ainsi que des agrégations annuelles composées de 15 habitations ou plus. Après 3800 BP, il y a absence de grandes agrégations telles qu’au pré-Dorset. Pendant le Dorset, les types de peuplements sont dominés par des maisons longues multifamiliales et des rangées de foyers, ainsi que par des sites de deux à quatre habitations. Il est possible qu’à l’époque du Dorset, les occupants vivaient surtout dans des habitations en neige placées sur la glace de mer en saison froide, tandis que les maisons longues et les rangées de foyers étaient installées sur la terre. Les caractéristiques architec- turales des maisons longues et des rangées de foyers indiquent que leur construction et leur utilisation avaient une raison d’être commune au sein de la région et apparemment dans l’Arctique canadien, mais leur emplacement et le moment de leur origine demeurent obscurs. D’après la datation au radiocarbone, la plupart des maisons longues et des rangées de foyers de l’ouest de l’île Victoria remontent au Dorset tardif, comme ailleurs, mais certaines dates indiquent que ces structures étaient utilisées dans l’ouest de l’Arctique canadien vers le Dorset moyen. La datation au radiocarbone du Dorset tardif effectuée dans la région chevauche la plus ancienne des dates les plus fiables établies relativement aux habitations de Thulé. Mots clés : Paléoesquimau, pré-Dorset, Dorset, Thulé, démographie, type de peuplement, longue maison, rangées de foyers

traduit pour la revue Arctic par Nicole Giguère.

1 Department of Anthropology, McGill University, 855 Sherbrooke Street West, Montreal, Quebec H3A 2T7, Canada 2 Corresponding authors: [email protected] and [email protected] 3 Geological Survey of Canada, 601 Booth Street, Ottawa, Ontario K1A 0E8, Canada, and Department of Geography, Memorial University of Newfoundland, St. John’s, Newfoundland A1B 3X9, Canada 4 Department of Archaeology, Memorial University of Newfoundland, St. John’s, Newfoundland A1C 5S7, Canada © The Arctic Institute of North America 168 • J.M. SAVELLE et al.

INTRODUCTION there is no compelling evidence that the early Paleoeskimo population decline was climate-induced. Rather, we sug- Savelle and Dyke (2002) documented Paleoeskimo demo- gested that on Victoria Island, the population crashes may graphic history on western Victoria Island (Fig. 1) on have been due to the overhunting of a critical resource, the basis of surveys in six regions: Page Point, Wood- muskoxen. Regardless of the reason for the boom-and-bust ward Point, Linaluk Island, south-central Prince Albert cycles, however, societies would necessarily have adjusted Sound, Innirit Point, and Cache Point (see Savelle and to them. Dyke, 2002:510 for specific locations on western Victoria Island). They identified a series of “boom-and-bust” population cycles from changes in the relative abundance of APPROACHES TO PALEOESKIMO SETTLEMENT dwelling features throughout the Paleoeskimo period (ca. ARCHAEOLOGY 4500 – 800/600 14C years BP; henceforth BP). Populations were at their highest during the early Pre-Dorset period. A Although several approaches might be followed, we rapid decline followed, beginning about 3800 BP, and by restrict our analysis to an examination of variability in site ca. 3200 BP, populations were at minimum levels or were sizes, site size frequencies, and feature types, which can be absent. Although a slight peak occurred in Early Dorset expected to reflect band sizes and fission-fusion behaviours, time at about 2200 BP, it was not until the latest stage of and ultimately, social relations. Band size and fission-fusion the Dorset period that population levels rose substantially. behaviour, in turn, are directly influenced by the nature of Nevertheless, these later increases do not appear to have the resource base (e.g., Binford, 1980, 2001). reached anywhere near the levels of the early Pre-Dorset In this paper, as in our previous studies cited above, we occupation. In this paper, following similar paleodemo- define a site as (a) an individual dwelling feature isolated graphic studies on Boothia Peninsula (Savelle and Dyke, from (and generally not within sight of) any others or (b) 2009), on King William Island and Kent Peninsula (Dyke an obvious grouping of two or more dwellings spaced a few and Savelle, 2009), and around the Gulf of Boothia (Dyke meters apart, at the same elevation or closely similar eleva- et al., 2011), we augment the treatment of the Victoria Island tions on adjacent beach ridges. The question of contempo- record by Savelle and Dyke (2002). We discuss changes in raneity of features is important, especially at larger sites, population dynamics by examining site sizes and dwelling but it is impossible to prove by direct dating. We deal with characteristics through the boom-and-bust cycles summa- this issue further below. rized above. In addition, we describe previously unreported Dorset longhouses and hearth-row sets and present new radiocarbon dates that may have significance for Dorset- DATING OF SITES Thule interaction. We use site elevation as an indicator of relative dwelling age because prehistoric and historic Eskimo site position- BACKGROUND ing tends to have been strongly correlated with contemporaneous sea levels (Meldgaard, 1960, 1962; McGhee, 1976, Data in the present study were derived from surveys 1979, 1981; Schledermann, 1978). The set of radiocarbon of six areas on western Victoria Island in 1999 and 2000 dates derived from various Paleoeskimo features and listed (Savelle and Dyke, 2002), from a survey of a seventh area in Savelle and Dyke (2002) is augmented here with addi- at Walker Bay on northwestern Victoria Island in 2001, tional dates from Walker Bay (Table 1). All dates reported from excavations at a longhouse – hearth row set at Innirit here and in Savelle and Dyke (2002) are consistent with the Point in 2000, and from faunal analyses in 2006 from the notion that elevation above sea level corresponds to approx- same feature. With the probable exception of Walker Bay, imate site age. Combining the two suites of dates, 62 of the these areas have experienced continuous isostatic rebound 71 dates are considered in this context. Of the remainder, since approximately 10 000 BP. Relict shorelines are pro- six dates on shell are excluded because they may date the gressively older upslope, with the oldest shorelines defin- beach rather than the occupation, whereas three dates are ing the limits of postglacial submergence (approximately excluded because of their association with features that are 50 – 150 m). Because of these circumstances, essentially the on bluffs (lookouts). For the remaining 62 dates, despite an entire occupational history is archived within the archaeo- overall east-west gradient in uplift rates that might produce logical record on the raised beaches. a 1 – 2 m difference in beaches of similar ages between The rapid increase in population following initial coloni- the eastern and western extremes of the study area, there zation noted above (Fig. 2) is consistent with the behaviour is a strong positive Pearson’s correlation between feature of all species, including humans, moving into previously age and elevation (0.768; p < 0.01). If these 62 samples are unoccupied environments, and subsequent crashes are divided according to materials, the strong correlations are likely consequences of over-population. We have discussed maintained. For “immediate use” materials, Salix sp. (wil- these cycles in Savelle and Dyke (2002, 2009), Dyke and low) charcoal and terrestrial mammal bone (n = 42), Pear- Savelle (2009), and Dyke et al. (2011). We concluded that son’s correlation is 0.772 (p < 0.001). For Picea sp. (spruce) PALEOESKIMO DEMOGRAPHY ON VICTORIA ISLAND • 169

Beaufort Sea BANKS ISLAND FIG. 2. Histogram of Paleoeskimo dwelling frequency by elevation (m asl) at Woodward Point. This pattern is consistent throughout our study area where Walker Bay we have essentially complete raised beach sequences (see Savelle and Dyke, 2002). Amundsen Gulf VICTORIA Holman Woodward Point the second population increase took place during Early to Prince Albert Sound

Wollaston Middle Dorset time, we include all Dorset sites in one cat- ISLAND Peninsula

Innirit Point egory, because the small sample size, especially of Early Cambridge Bay Dorset sites, limits the usefulness of any comparison of these sites with sites dating to other intervals. Those few sites for which appropriate data were not collected are not included in the demographic analyses. However, our approach does allow for the recognition of patterns in residential site sizes, which is appropriate for NUNAVUT broad-scale demographic analyses and offers more com- N.W.T. prehensive insights into demographic processes than simple feature or site counts alone. Dwelling units recorded herein include well-defined 0 200 axial features (midpassages) with or without peripheral km gravel or stone rings, scattered stone arrangements typically with associated sod overgrowths, well-defined box FIG. 1. Map of the study area, showing locations of the three longhouse sites hearths typically containing boiling stones, linear hearth- discussed in the text (Innirit Point, Woodward Point, Walker Bay) and the present settlements of Holman and Cambridge Bay. row sets, and longhouses (Fig. 4). With the possible exception of the longhouses, no Paleoeskimo features were recorded that suggest anything other than single seasonal charcoal (n = 20), Pearson’s correlation is 0.780 (p < 0.001). occupations. Furthermore, of the hundreds of features Although dates on driftwood provide maximum age esti- recorded at Paleoeskimo sites, caches are rare and small mates, the great abundance of driftwood in the region (less than 0.5 m) piles of stone slabs. For the purposes of (Dyke and Savelle, 2000) likely ensured that the Paleoeski- feature counts, we treat each hearth row, hearth-row set, mos availed themselves of fresh wood rather than collecting and longhouse as an individual feature representing as subfossil wood from raised beaches. If so, the overestima- many social units as there are hearths or hearth sets. We tion of the age should be at most a few centuries. do not assume contemporaneity of these features. Doing so would lead to interpretations of band aggregations much larger than any ethnographically documented for mobile RESULTS AND DISCUSSION Arctic hunter-gatherer societies.

Within the seven study areas for which we have infor- Early Pre-Dorset mation on site elevation and site size, 745 Paleoeskimo dwelling features occur at 77 sites. On the basis of the pop- At sites of early Pre-Dorset age, single or few dwellings ulation boom-and-bust cycles of Savelle and Dyke (2002), are most common. Sites of 1 – 2, 5 – 6, 11 – 15, and 19 – 20 we divided that sample into three age categories: Early dwellings likely represent group fission-fusion cycles Pre-Dorset (initial immigration and population expan- (Fig. 5). Amongst historically documented mobile Inuit, sion), Middle and Late Pre-Dorset (population decline), and household size varied, but most households numbered four Dorset. Figure 3 presents a detailed map of a typical beach or five individuals. Birket-Smith (1929) and Rasmussen sequence with Early Pre-Dorset to Late Dorset dwellings (1931) provide details of family size for Caribou Inuit and and—below 7 m—Dorset and Thule dwellings. Although Netsilik Inuit, respectively, and Binford (2001:296 – 297) 170 • J.M. SAVELLE et al.

TABLE 1. New radiocarbon dates on Paleoeskimo features from Walker Bay and Cape Back and other Paleoeskimo dates referred to in the text. The last column gives the fractional area under the calibration curve for the stated range.

Lab code Field site (Borden code) Material elevation (m asl) 14C Age (δ13C) Calib 6.0 2-sigma range BP Area

Walker Bay: AA-46729 01-68 (OjPt-5) Salix charcoal 3 1995 ± 60 (-26.7) 1734 – 2043 1 AA-46730 01-68 (OjPt-5) Picea charcoal 3 2050 ± 62 (-23.8) 1874 – 2154 0.986 AA-46731 01-70 (OjPt-4) Picea charcoal 3 1235 ± 73 1049 – 1291 0.926 AA-46732 01-71 (OjPt-3) Picea charcoal 1 – 1.5 1375 ± 6 (-25.9) 1168 – 1412 0.987 AA-46733 01-71 (OjPt-3) Picea charcoal 1 – 1.5 1064 ± 50 (-24.0) 0905 – 1081 0.977 AA-44352 01-71 (OjPt-3) Picea wood 1 – 1.5 1267 ± 38 (-25.8) 1122 – 1286 0.947 AA-46734 01-71 (OjPt-3) Picea charcoal 1 – 1.5 1004 ± 54 (-26.5) 0789 – 1007 0.957 Cape Back: TO-8149 98-48 (OaPo-1) Driftwood charcoal 10.5 3250 ± 80 3334 – 3645 0.975 Previously Published: AA-41516 99-160 (NhPl-1) Salix charcoal 7 1925 ± 71 (-26.4) 1701 – 2042 1 AA-41512 99-160 (NhPl-1) Rangifer or Ovibos bone 7 1893 ± 42 (-20.2) 1720 – 1904 0.962 TO-8532 99-160 (NhPl-1) Picea and Salix charcoal 7 2100 ± 60 1925 – 2183 0.880 AA-41503 00-121 (OdPc-10) Burnt moss 8.5 1575 ± 40 (-26.4) 1379 – 1544 1 AA-41502 00-121 (OdPc-10) Salix charcoal 5.5 1162 ± 39 (-26.4) 0973 – 1176 1 AA-40580 00-121 (OdPc-10) Picea charcoal 4.5 1443 ± 42 (-25.7) 1292 – 1402 1 AA-40857 00-121 (OdPc-10) Picea charcoal 4.5 1318 ± 35 (-25.7) 1220 – 1298 0.726 AA-40860 00-130 (OdPc-13) Salix charcoal 4 – 4.5 0816 ± 37 (-26.9) 0677 – 0789 1 AA-41504 00-130 (OdPc-13) Salix charcoal 4 – 4.5 0619 ± 54 (-27.0) 0537 – 0668 1 AA-41506 00-135 (OdPc-8) Salix charcoal 9.5 0806 ± 39 (-25.2) 0674 – 0785 1 GSC-6346 (OaPp-1) Picea and Salix charcoal 9.5 0650 ± 50 (-26.33) 0558 – 0668 1 summarizes family sizes of Inuit in general. Thus, and Thus, the patterns exhibited on Victoria Island are consist- following others (e.g., Maxwell, 1985:98; McGhee, ent with that behaviour. As we have discussed previously 1996:123 – 124), we suggest that for much of the year, single (Dyke and Savelle, 2009; Savelle and Dyke, 2009; Dyke families or groups of two or three families lived and hunted et al., 2011), we are not assuming simultaneous occupation together, probably moving frequently. However, as noted for all dwellings at every site, but the suggestion that larger by Jenness (1922), for much of the fission part of the cycle sites represent larger gatherings is consistent with hunter- during historic times, groups of 12 – 20 individuals were gatherer behaviour. The peaks at 11 – 15 and 19 – 20 dwell- also common. Accordingly, the sites with five or six dwellings represent aggregations within the size of historic Inuit ings may indicate micro-bands in this size range or perhaps gatherings, and even the largest sites (up to 40 dwellings) slightly larger. The larger sites, on the other hand, are likely are consistent with historical Inuit group aggregations. the result of seasonal band aggregations (e.g., Binford, 1980, 1982; Kelly, 1995). Historic maximum band aggregations Middle and Late Pre-Dorset amongst mobile groups in the Arctic ranged up to about 150 (Binford, 2001:251). Progressively larger sites, especially The Middle and Late Pre-Dorset pattern of site sizes those with 11 – 15 and 19 – 20 features, and possibly those (Fig. 5) differs considerably from the Early Pre-Dorset pat- with 22 – 26 features, are consistent with amalgamation of tern. Although sites with one or two features remain the five- to six-household micro-bands into increasingly larger most common, the household amalgamation sites, espe- units until macro-band size is attained. The large cluster of cially in the five- to six-household range, are lacking, and dwellings on the 12 – 13 m beaches at Innirit Point (Fig. 3) no sites have more than eight dwellings. Because the Mid- may represent such an amalgamation. These aspects of the dle and Late Pre-Dorset period is longer than the Early Pre- early Pre-Dorset settlement pattern are consistent with that Dorset period (1000 vs. 700 years), and because our survey of historically documented mobile Arctic hunter-gatherers. intensity was similar on all beach sequences, this difference Given the lack of any evidence of significant food storage, is not likely due to our failure to locate larger Middle and the pattern fits well within Binford’s (1980) foraging type of Late Pre-Dorset sites. Instead, the different pattern suggests settlement system: that is, a system used by highly mobile a substantially different settlement system, whereby mini- societies caching little or no food. mum unit sizes of one to two households were presumably As noted above, the question of contemporaneity of fea- maintained for much of the seasonal round, and amalgama- tures is important. It is impossible to prove by dating meth- tion tended to be restricted to micro-bands. The practice of ods, and as far as we can see, the problem is intractable. macro-band gatherings of the Early Pre-Dorset societies However, we see no reason to doubt that most features at had evidently been abandoned. The lack of macro-band a site were simultaneously occupied. Because essentially sites may be partly a function of lower regional population all hunter-gatherer groups exhibit seasonal fission-fusion size. However, if key prey resources had declined signifi- behaviour (Kelly, 1995), one kind of evidence of a band cantly, the smaller Middle and Late Pre-Dorset bands might aggregation to be expected is a site with many dwellings. also reflect the cost-benefit ratio of varying group sizes. PALEOESKIMO DEMOGRAPHY ON VICTORIA ISLAND • 171

FIG. 3. Sites on a raised beach sequence at Innirit Point. Feature 160-1 is a Dorset longhouse or triple hearth-row set. Dwelling features at and above 12 m are of Pre-Dorset age. Dwelling features at about 6 m and lower are of mixed Dorset and Thule origin, and many of the higher-level caches are of Thule age. 172 • J.M. SAVELLE et al.

FIG. 4. Paleoeskimo dwelling features. a) Longhouse or triple hearth row at Innirit Point, view east; b) same feature, view south; c) longhouse at Woodward Point, view southeast; d) hearth-row sets 1 and 2 at Woodward Point, view west; e) Late Dorset hearth at 4 – 4.5 m asl at Woodward Point, chert cores beside notebook; f) another Late Dorset hearth at 4 – 4.5 m asl at Woodward Point, chert cores beside notebook; g) longhouse at Walker Bay, view east; and h) hearth rows 1 and 3 at Walker Bay, view east and a row of solitary boulders south of hearth row 1. PALEOESKIMO DEMOGRAPHY ON VICTORIA ISLAND • 173

are longhouses or hearth-row sets. Furthermore, one of the six-unit features and one of the three-unit features are also hearth-row sets. Building materials also changed at this time. Whereas flagstones were used almost exclusively in constructing Pre-Dorset dwellings, boulders of up to 1 m diameter and heavy blocks were commonly used in constructing longhouses and hearth rows. Although we do not know how many hearth-row sets at any site may have been occupied simultaneously or if hearth-row set and adjacent longhouse occupations were always contemporaneous, we follow Damkjar (2000) and Appelt and Gulløv (1999) in treating each longhouse and hearth-row set as representing a separate occupational episode. We use the term longhouse to designate solidly constructed features with stone walls and interior compartments (Fig. 4), whereas hearth-row sets lack solid walls and occur as single lines of hearths or as two or more parallel lines of hearth sets (Fig. 4). Some hearth-row sets have well-defined rectangular flagstone perimeters, such as the one at site 99-160 (see below). There is little doubt that longhouses and hearth rows represent band—probably macro-band—amalgamations, and several reasons for their development have been inferred. Plumet (1982) suggested that they represent regional macro- band occupations during the season after nighttime dark- ness had resumed but before occupation of winter houses and that they had not been adopted from other longhouse- using cultures. Schledermann (1990) suggested that they represent communal cooking and sharing of food; McGhee (1996) and Appelt and Gullov (1999), that they functioned as centres for communal socializing, dancing, contests, and ritual celebrations; Damkjar (2000), that they represent responses to subsistence resource scarcity by fostering communal sharing within an expanded household; Murray (1999), that they symbolically expressed a sense of commu- nity or territory amongst relatively mobile Dorset groups; Friesen (2007), that they expressed social egalitarianism; and Park (2003), that they were constructed by a few individuals only, the construction emphasizing a worldview of linearity. It is likely that longhouses and hearth rows served FIG. 5. Histograms of site sizes for Early Pre-Dorset, Middle and Late Pre- one or more of these functions. Dorset, and Dorset periods. Corresponding with the development of these new macro-band features during the Dorset period is a reduc- Specifically, as suggested by Binford (2001:242), when prey tion in the frequency of single-family dwellings and micro- resource abundance decreases, greater net benefit is associ- band settlements of more than six dwellings (Fig. 5). The ated with smaller group sizes, because the energy required most parsimonious explanation for this pattern is that many to move from one locality to another (as required of groups of these settlements were then located on the sea ice. In this that cache little or no food) is less for smaller groups than regard, McGhee (1996:118) has summarized the marine larger groups. mammal-oriented subsistence economy of Dorset culture, noting that “more extensive sea ice and a longer period dur- Dorset ing which this hunting platform was available may well have produced a more efficient and productive economy” Although Dorset sites are few, most of the dwelling fea- relative to earlier Paleoeskimo cultures. Furthermore, ice- tures are similar in form to those of earlier periods and adapted equipment, such as ice creepers and snow knifes, also similar in the lack of any semisubterranean (cold- has been found in Dorset sites, suggesting that the Dorset season) houses. The prominence of 10 – 11 and 14 – 15 Dor- possessed a snow-based technology that included the con- set household sites (Fig. 5) provides the main contrast to struction of snow dwellings (Maxwell, 1985). This technol- the earlier periods. All Late Dorset sites of 10 or more units ogy, which included the sea mammal oil lamp—a necessity 174 • J.M. SAVELLE et al. for the heating of snow-based dwellings—may indicate that Neoglacial cooling had reached the point at which the increase in the duration of sea ice, and hence ringed seal availability, forced or enticed a fundamental economic reorientation. Alternatively, the mere advent of the oil lamp may have rendered the snow house a viable long-term dwelling. The putatively correlative longhouse appeared no earlier in cold North Greenland than in warmer regions (Damkjar, 2000 and below), which hints at a lack of climate forcing.

VICTORIA ISLAND LONGHOUSES AND HEARTH-ROW SETS

One facet of Dorset longhouses and hearth rows that has received little attention, except in studies by Appelt and Gulløv (1999) and Friesen (2007), is the social composition of the groups using them. Although these groups are generally believed to represent macro-band aggregations and various social and ritualistic functions, little else has been inferred. Part of the problem has been that, while inter- site variability in size of longhouses has been investigated (Damkjar, 2000:174), intra-site descriptions of larger sites are rare. However, as demonstrated by Appelt and Gulløv (1999) and Friesen (2002, 2007), such data can reveal important aspects of Dorset social behaviour. In our study area, three sites with longhouses and hearth-row sets, 01-71 (OjPt-3), 00-121 (OdPc-10), and 99-160 (NhPl-1), allow us to build on those previous investigations. FIG. 6. Sketch maps of the longhouse and hearth-row sites east of Woodward Point and at Walker Bay. Boulders within longhouses are schematic only. The Site 00-121 (OdPc-10) diagram for site 00-121 is not to scale. Site 00-121 east of Woodward Point (Fig. 1) consists of one longhouse, originally reported by McGhee (1971), (Fig. 6). The set is 23.95 m long and consists of three par- and three hearth-row sets (Fig. 6). The longhouse (Fig. 4) allel rows of 15 hearths each, spaced 4 – 5 m apart, with is located at 8.5 m elevation about 230 m inland from the the hearths in each row positioned opposite hearths in the hearth rows. It is 31 m long and varies in width from 5.35 to neighbouring row. As in set 1, the seaward row contains 5.9 m. It contains an axial alignment of circular to irregular hearths that are more prominent and larger. boulder structures that appear to represent living areas or Set 3 is approximately 200 m east of the first two sets. midpassage hearths, of which there are 14 or 15. An exter- This set is curved and it obliquely crosses one raised nal box hearth measuring 40 × 55 cm is located 3 m north of beach ridge and the swale to the seaward. The lower end the longhouse. About 200 m north is a square Paleoeskimo of the feature is 5.5 m above the high tide line. The curved dwelling feature measuring about 3 × 3 m on coarse beach length of the set is 24.55 m, and it varies in width from 2.8 gravel at 16.5 m elevation. This is the only feature, other to 3.0 m. The set consists of two parallel lines of hearths, than the hearth-row sets, in the vicinity of the longhouse. with 14 large hearths in the seaward row and 12 smaller Hearth-row set 1 is at 4.5 m elevation, oriented along a hearths in the landward row. The hearths are in opposite raised beach of coarse gravel. The set consists of two paral- pairs, except that two hearths are missing in the landward lel hearth rows, one defining the seaward side, the other the row. One of these was the corner hearth at the west end. inland side, with regularly spaced hearths. The 15 hearths The other, the eighth hearth from the west, has been re-used in each row are distinctly paired; that is, they are positioned as an inukshuk in a caribou drive line that runs through the opposite to each other. The seaward hearths are large and feature. the landward hearths are smaller and less conspicuous. The In addition to the features described above, approxi- set is 27.75 m long, and the rows are 3.2 m apart at the east mately 3 km southwest of site 00-121 is a short late end and 3.5 m apart at the west end. Paleoeskimo hearth-row set (site 00-139, OdPc-21) located Set 2 is 1 m inland of set 1, is also oriented along a raised on a local strip of fine beach gravel surrounded by coarse beach of coarse gravel, and overlaps set 1 in length by 6 m gravel at 6.5 m elevation. The structure has six pairs of PALEOESKIMO DEMOGRAPHY ON VICTORIA ISLAND • 175 hearths, and the seaward hearths are larger than the land- set 5. This set is either an external hearth row associated ward ones. Numerous chert cores and flakes are clustered with the longhouse or it is the axial feature of an older struc- at the hearths. The feature is 8.45 m long, close to the short ture that predates the longhouse. The geometric configura- end of longhouses reported by Damkjar (2000), and it tion suggests that the first interpretation is more probable. ranges in width from 3.55 to 3.7 m. Set 5 is composed of two parallel hearth rows with large seaward hearths and smaller landward hearths located Site 01-71 (OjPt-3) 25 m northwest of the longhouse. Some landward hearths have probably been removed in part by Thule people for Site 01-71 is located on a high, narrow raised beach construction of meat caches nearby. The seaward row has ridge 1 m above high tide in a cove along the north shore 11 hearths and the landward row has 11 small or partial of Walker Bay about 7 km north of Pemmican Point and hearths. The set is 18.7 m long. The row of larger hearths is about 15 km east-northeast of Berkeley Point (Figs. 1 and 1.8 m wide. The gap between the two rows ranges from 0.7 6). The raised beach is the first behind the active one, and to 1.2 m, and the inland row is about 1 m wide. it is bracketed by long swale ponds on both the inland and Row 6 is a single row of 11 evenly spaced boulder seaward sides. The site consists of one longhouse and six hearths, all large. The largest boulders are 30 – 40 cm. An hearth-row sets along about 150 m of beach ridge, all ori- ice-wedge trough 1.5 m wide runs across the feature. The ented parallel to the beach (Fig. 6). feature is 14.8 m long on one side of the trough and 1 m The longhouse is similar to that at site 00-121, being con- long on the other (total 15.8 m). structed from large boulders up to 1 m in diameter, with average diameter of 0.5 m (Fig. 4). The longhouse trends Site 99-160 (NhPl-1) parallel to the beach ridge a little seaward of the ridge crest. It is 21 m long, with a consistent width of 6.5 – 6.6 m. How- Feature 99-160-1, located at Innirit Point (Figs. 1, 3, ever, its interior is very disturbed, with three nearby tent and 4), was excavated in the summer of 2000. The feature rings and two boulder caches—all probably Thule—appar- yielded mainly pieces of worked chert, wood and bone deb- ently constructed from boulders removed from it. Deter- itage, charcoal and burnt bone, and only two dozen or so mination of the number of original compartments was formal tools (microblades, utilized flakes, end scrapers, a thus difficult, but it appears to have originally contained bifacial tool, a bone needle), none of which are diagnostic of between 9 and 11. A hearth row parallel to and immediately any particularly part of Dorset time. The feature measures adjacent to the longhouse is of identical length and contains about 9 × 4 m and seems to consist of three rows of hearths. 11 hearths. It is therefore likely that the longhouse origi- The central row is the most prominent in both the use of nally contained 11 compartments. slightly larger flagstones and the concentration of anthro- Hearth-row set 1, about 30 m southeastward along the pogenic vegetation along it (Fig. 7). In addition, all three beach ridge from the longhouse, consists of an inland row rows occur as nearly continuous linear arrangements rather containing 11 hearths over a length of 18 m. The seaward than as series of individually isolated hearths, so that, side of this set is a boulder line parallel to and 3 – 4 m from together with scattered but nevertheless discernable flag- the hearth row, with 12 boulders but lacking hearths. It stone ends, the whole forms a rectangular plan with a cen- starts 11 m east of the inland row and is 13 m long. tral axial feature. Overall, this pattern is more reminiscent Set 2 consists of two parallel and paired hearth rows of a longhouse with axial living compartments than of rows ranging between 4.7 and 5.5 m apart. The seaward row is of isolated hearths. The feature is unique amongst the long- 15 m long, but a probable former corner hearth is marked houses and hearth-row sets described in this paper and pos- by a gravel depression 1.5 m farther along this line. The sibly elsewhere. It is morphologically intermediate between inland row is 17.8 m long, extending past the probable miss- longhouses and hearth rows. ing hearth in the seaward row by 1.5 m. The inland row The hearth groupings in this feature are not as regular contains 11 hearths. The seaward row contains 10 hearths as those at sites 00-121 and 01-71, but patterns are discern- and the depression probably representing an eleventh. able. In Figure 7, structural rock and charcoal and burned Set 3 comprises two parallel and paired hearth rows bone locations define the clearest hearths. In addition, two spaced 3.2 – 3.23 m apart. It is 3 m west of set 2 and 2 m upright stones with associated vegetation in the southern inland of set 1. The seaward row is 17 m long, and the land- row—in the gap between three delineated hearths to the ward row 16.1 m long. The seaward row contains 11 distinct east and five to the west—are probable hearth remnants. hearths, and the inland row, 11 smaller hearths. Six of the defined hearths form closely spaced pairs, four Set 4 is a disturbed hearth row on the crest of the beach at the west end of the middle row, two in the southern row. ridge 3.5 m inland from the longhouse. It is parallel to and These presumably represent reuse of hearth areas. If so, the extends the full length of the house. The row contains 7 southern and middle rows each contain nine hearths. The distinct hearths and 4 indistinct stone features, probably northern row is much weaker, but it is the same length as disturbed hearths. Thule tent rings and meat caches in the the other two, which suggests that, if indeed it is a hearth immediate vicinity were probably constructed from boul- row (and the one identified hearth supports this interpreta- ders removed from the longhouse, this hearth-row set, and tion), then most of the row of flagstones represents hearths 176 • J.M. SAVELLE et al.

FIG. 7. Field drawing of the Innirit Point longhouse – hearth row set. One-meter grid points are indicated by +. as well. A weak landward hearth row would be consistent have fluctuated annually in size, as they did historically, with the pattern noted in the double and triple hearth rows it is highly unlikely that the social groups using the long- at Woodward Point and Walker Bay. houses and hearth rows were simply residential nuclear or The faunal remains from this feature (Table 2) are few extended family units. Rather the bands using these sites and small. Identified species are ringed seal, caribou, eider appear to have been more highly socially structured than duck, and collared lemming, an assemblage that does not were earlier Paleoeskimo groups. refute the suggestion by Damkjar (2000, 2005) that long- Others have also noted consistent numbers of hearths houses and hearth rows probably represent warm-season amongst hearth-row sets and between longhouses and sites. Given the high proportion of seal, both as number of hearth rows. Appelt and Gulløv (1999) noted a consist- identified specimens (NISP) and as minimum number of ency of 8 – 10 hearths or hearth pairs amongst the eight individuals (MNI), and the presence of apparently young hearth rows at the Hatherton Bay longhouse site in north- as well as adult eider duck, the feature was probably occu- west Greenland. In the longhouse itself are eight central pied during the early summer. However, surface sites such hearth pits, but “one or two pits are missing in the southern as this one are subjected to extensive decomposition pro- half” (Appelt and Gulløv, 1999:153). Similarly, at the long- cesses, and the denser seal bones, especially humeri and house site at Brooman Point, Park (2002:19) illustrates three femora (Lyman, 1994), may be overrepresented. hearth rows. Two are single rows, each containing 7 hearths (Park numbers only 6 in hearth-row 2, but there appears to be a seventh at the east end). The other is a double hearth LONGHOUSES, HEARTH-ROW SETS, row, containing eight hearth pairs, with the southern (sea- AND LATE DORSET SOCIAL STRUCTURE ward) row consisting of small hearths and the northern row of large hearths. The longhouse at that site is considerably The consistent number of hearths within hearth disturbed, as later Thule occupants used many of the heav- rows (or row sets) and longhouses at each site cannot be ier stones to construct a grave, which precludes discerning mere happenstance. Rather, it suggests a consistent pur- internal compartments. Finally, Schledermann (1996:98) pose behind their construction and use, unlike the pre- noted similar associations of large and small hearths in sumed relatively flexible fission-fusion behaviour of hearth-row sets on Ellesmere Island. earlier Paleoeskimo groups and of many historic Arctic The change from individual tents characteristic of the Inuit societies. Given that we can expect macro-bands to earlier Paleoeskimo periods to longhouses and hearth rows PALEOESKIMO DEMOGRAPHY ON VICTORIA ISLAND • 177

TABLE 2. Faunal remains, Innirit Point longhouse – hearth row suggests that Dorset social organization was probably far set. more complex than previously realized.

mnI2 Species NISP1 % NISP /Age Group mnI % MNI HISTORY OF LONGHOUSE – HEARTH ROW Ringed seal 108 8 SET COMPLEX Ringed seal? 3 Ringed seal infant 1 1 Ringed seal total 112 45.2 9 60 The early dates for the hearth-row set at Innirit Point Caribou 33 1 (Savelle and Dyke, 2002; Table 1) have implications for Caribou or muskox 44 the history of development of the longhouse – hearth row Caribou-muskox total 77 31.1 1 6.7 Eider duck 15 2 complex. The place of origin of the Dorset longhouses is Eider duck chick 2 1 still in doubt. Damkjar (2000:178) suggests they may have Eider duck total 17 6.8 3 20 originated in Ungava, where they are particularly abundant Bird 25 10.1 Collared lemming 14 2 (Plumet, 1982), although with semisubterranean precur- Lemming? 3 sors in Foxe Basin, and Murray (1999:476 – 477) likewise Lemming total 17 6.8 2 13.3 suggests a Foxe Basin origin from semisubterranean pre- Total 248 100 15 100 Unidentified bones 167 cursors. McGhee (1996:176, 210) has listed a number of Mollusc shell fragments 4 possibilities, including borrowing of the concept from the Cree of the northeastern forests (see however, Plumet, 1 Number of identified specimens. 1982) or independent invention. The style is believed to 2 Minimum number of individuals. have spread subsequently to the Canadian Arctic Archi- pelago during Late Dorset time, approximately 1200 – 1300 represents a change from use of private spaces to use of BP (Schledermann, 1990). The dates available prior to this one large public space during macro-band aggregations. study (Damkjar, 2000:175) support that chronology. Most According to Whitridge et al. (2001), these new public fea- of our dates from the Woodward Point site, specifically all tures represented a physical agency for ensuring coordi- three dates on the hearth row sets at 4.5 – 5.5 m, and the nated social action, which previously had to be coordinated dates on the Walker Bay hearth-row sets are also consistent and negotiated during each aggregation. In addition, if the with that chronology (Table 1). However, the three Innirit longhouse builders were the first to live in snow-house Point hearth-row dates (1893 ± 42 BP (AA-41517), 1925 communities on the sea ice, then they were also the first ± 71 BP (AA-41516), 2100 ± 60 BP (TO-8532)) based on that had to come ashore in the spring. Accumulated seal a caribou or muskox long bone fragment, willow charcoal, meat may have allowed them to enjoy the transition from and a willow-spruce charcoal mixture, respectively, and spring to summer as a group. Alternatively, or additionally, a date of 1575 ± 40 BP (AA-41503) on burnt moss from a these sites may have been occupied during a waiting period hearth beside the Woodward Point longhouse at 8.5 m, indi- prior to fall sea ice freeze-up. cate an earlier development. We have no reason to suspect In the case of double and triple hearth-row sets, there is that at least the first two dates from Innirit Point and the also a strong consistency between sites in hearth structure. Woodward Point date are not valid measures of the occupa- Specifically the pairing of strongly built and lightly built tion dates. Accordingly, we suggest that by the approximate hearths in row sets prevalent on Victoria Island is also seen beginning of the Middle Dorset period (ca. 2000 – 1500 BP), at the Hatherton Bay and Brooman Point sites. At Hather- the longhouse – hearth row complex was already present in ton Bay, two of the hearth-row sets (1a and b, 5a and b) are the western part of the Canadian Arctic Archipelago. The nearly identical to the Victoria Island large and small sets hearth rows at Woodward point are apparently several cen- (Appelt and Gullov, 1999:33), and at Brooman Point, the turies younger than the longhouse and are not within sight double hearth-row set also contains noticeably large and of it. small pairs. Understanding the significance of this pattern- Savelle and Dyke (2009) discussed exceptionally large ing requires a much more thorough investigation than ours, Paleoeskimo dwellings on western Boothia Peninsula as but given the widespread distribution of these structures possible precursors of the longhouses. Of the 481 dwell- (throughout Arctic Canada and northwestern Greenland), ing features there, the five largest—ranging from 32.0 a strong element of symbolism and extensive inter-regional to 48.7 m2—occur at small sites within elevations cor- contact must have been involved. responding to the Early Dorset period. Two of these fea- As noted by Damkjar (2000:178), “we will be rethinking tures are isolated, one occurs with a smaller dwelling, the development of Dorset longhouses...for many years to and the remaining two with two other small dwellings. come.” There are many directions future studies could take, In addition, at the Dorset site of Alarnerk in Foxe Basin, but the evidence from Victoria Island suggests that long- Savelle and Dyke (unpubl. data) re-examined Meldgaard’s houses heralded important changes in social behaviour dur- (1960, 1962) “large houses.” These rectangular, semisub- ing macro-band amalgamations. Regardless of where these terranean features occur at 19 m elevation and measure changes may have originated, the nature of the changes 15 × 8 m and 12.3 × 7.8 m. The sizes and dug-in nature 178 • J.M. SAVELLE et al. indicate communal winter dwellings, and they are pre- or that it occurred as late as the 13th century (McGhee, sumably the possible longhouse precursors referred to by 2000; Friesen, 2004), these dates suggest that a remnant Damkjar and Murray cited above. Our date of 1690 ± 15 Dorset population existed on western Victoria Island dur- BP (UCIAMS-53038) on caribou bone from the floor of ing the Thule migration. Our dates on Thule sites in the the larger dwelling suggests that dwellings comparable area (Table 3), supplemented with dates from Woodward in size to longhouses in other regions were also in use in Point mentioned in Outridge et al. (2009), also indicate the Foxe Basin region by Middle Dorset time. The differ- that there may have been temporal and geographic overlap ence in season of occupation may reflect winter surpluses in the region between Thule and Paleoeskimo. Overlap is based on walrus in Foxe Basin, whereas in other regions less easy to demonstrate in the northern part of the region, the communal occupations were short-lived and presum- where the older dates are on driftwood used as structural ably based on less abundant resources. Nevertheless, warm- members in Thule semisubterranean dwellings. However, season (i.e., morphologically typical) longhouses are several of the dates on terrestrial mammal bones from the unknown in the Foxe Basin region, and Dorset winter com- Thule houses at Woodward Point are earlier than the young- munal houses are unknown on western Victoria Island. est Dorset date mentioned above. The new dates reported here may simply indicate an ear- The dates from the large Thule site at Lady Franklin lier spread of the “Dorset communal dwelling” complex out Point, southwestern Victoria Island, may be significant in of Foxe Basin or significant influences on its development this regard. The dated materials derive from a group of much farther west. 19 semisubterranean dwellings at the 4 m level and two semisubterranean dwellings at the 5 m level. The site has been badly disturbed, and the dated material from the 4 m LATE DORSET-THULE INUIT CONTACT houses, except for the structural wood post, was collected from surface debris, although the whale rib fragment may Several radiocarbon dates resulting from our research have originally been structural as well. For the 5 m houses, have implications for the question of contact between Late the wood post was structural, and probably also the whale Dorset and eastward migrating Thule Inuit. Until recently, rib, judging by Taylor’s (1972) diagram of the dwellings. little beyond circumstantial evidence suggested contact. We dated a range of materials (caribou, bear, dog, and This evidence includes overlapping radiocarbon dates whale bone, and wood) in part to assess what the local res- between Thule and Late Dorset, similarities in some arti- ervoir effects might be. Although these effects are evident, fact forms, and the stratigraphic context of several Thule the caribou antler date (CAMS-66369) is again earlier than artifacts within Dorset features. Although some argued our latest Dorset dates. This site remains poorly dated, and against this evidence, ancient DNA studies (Hayes et al., its large size and location at a regionally important caribou 2005) have now conclusively established that the Sadler- crossing indicate that further dating is warranted to see if miut Inuit of Southampton Island, who became extinct as its earliest occupation dates to earliest Thule time. a result of introduced disease at the beginning of the 20th Overall, the study region represents the first point in century, were a product of interbreeding between Dorset Canada of Paleoeskimo-Thule contact for which the full and Thule populations (see Rowley [1994] for a detailed archaeological record during the contact period is extant, summary of the history of research on the Sadlermiut). because coastal areas farther west have been submerging The question, then, is not whether Dorset-Thule con- throughout the middle and late Holocene. Accordingly, tact occurred, but when and where, outside of Southamp- Victoria Island has high potential for the testing of various ton Island, it may also have taken place. Three of our models of the nature of this contact (see e.g., Friesen, 2000). Paleoeskimo dates, all on willow from Woodward Point (Table 1), are relevant in this context. Dates of 619 ± 54 BP (AA-41504) and 816 ± 37 BP (AA-40860) were obtained COMMENT AND CONCLUSIONS from hearths at 4.0 – 4.5 m elevation (Savelle and Dyke, 2002). Both hearths were similar in shape and size (about Because our study is primarily based on regional sur- 1.0 × 0.5 m), and the hearth producing the earlier of the two veys and radiocarbon dating, our interpretations have nec- dates had 14 chert cores and 5 large chert flakes in asso- essarily focused on regional expressions of Paleoeskimo ciation. The third date, 806 ± 39 BP (AA-41506), is from occupation and demographic history. Accordingly, the con- a hearth located between two tent rings, one containing clusions summarized below might best be considered as a probable midpassage, with chert flakes in association a preliminary framework, or connected hypotheses, to be (Savelle and Dyke, 2002). In addition, we have previously tested through additional surveys and site-specific studies. reported a date of 650 ± 50 BP (GSC-6346) on driftwood and willow charcoal from a Paleoeskimo midpassage near 1. During periods of high population levels, the fission- Cape Baring on the northwest corner of Wollaston Penin- fusion settlement cycle emphasized variously sized sula (Dyke and Savelle, 2000). macro-band aggregations, presumably as a response Whether one accepts that the initial Thule migration into to abundant muskoxen or other resources. In contrast, the area dates to the 11th century (e.g., Morrison, 2000) PALEOESKIMO DEMOGRAPHY ON VICTORIA ISLAND • 179

TABLE 3. New radiocarbon dates on Thule sites, western Victoria Island. Only terrestrial dates are calibrated. The last column gives the fractional area under the calibration curve for the stated range.

Lab code Field site (Borden code) material Elevation (m asl) 14C Age (δ13C) Calib 6.0 2-sigma range BP Area

Berkeley Point Area: AA-46729 01-68 (OjPt-5) Salix charcoal 3 1995 ± 60 (-26.7) 1734 – 2043 1 GSC-6603 01-30 H. 2 (OjPu-2) Picea wood 2 600 ± 50 (-25.92) 543 – 652 1 GSC-6599 01-30 H. 3 (OjPu-2) Driftwood 2 900 ± 90 (-25.99) 732 – 920 1 GSC-6601 01-41 H. 1 (OkPv-2) Larix wood 3 1010 ± 50 (-25.29) 905 – 969 0.939 AA-46723 01-42 (OkPv-3) Salix charcoal 15 392 ± 54 (-28.5) 315 – 516 1 AA-46726 01-52 (OlPv-1) Salix charcoal 2.5 164 ± 42 (-25.9) 0 – 289 1

Cape Ptarmigan: GSC-6605 01-12 (OgPt-2) Picea wood 5 570 ± 50 (-24.18) 531 – 642 1

Woodward Point: GSC-6534 00-133 (OdPc-16) Picea wood 10 – 10.5 960 ± 50 (-26.27) 795 – 929 1 AA-58937 00-133 (OdPc-16) Ovibos moschatus bone 10 – 10.5 663 ± 46 (-21.33) 552 – 679 1 AA-58938 00-133 (OdPc-16) Rangifer tarandus bone 10 – 10.5 644 ± 43 (-20.88) 551 – 670 1 AA-58939 00-133 (OdPc-16) Rangifer tarandus bone 10 – 10.5 649 ± 43 (-20.73) 552 – 672 1 AA-58940 00-133 (OdPc-16) Ovibos moschatus bone 10 – 10.5 589 ± 42 (-22.80) 533 – 655 1 AA-58941 00-133 (OdPc-16) terrestrial mammal bone 10–10.5 654 ± 41 552 – 674 1

Innirit Hills: TO-8536 99-181 (NhPk-2) Picea charcoal 3.5 1100 ± 60 925 – 1172 1 CAMS-65201 99-210 (NhPl-11) Charcoal 2 870 ± 40 695 – 907 1

Lady Franklin Point: CAMS-66369 99-225 H. 21 (NcPd-2) Rangifer tarandus antler 5 660 ± 30 (-19.3) 558 – 672 1 GSC-6422 99-225 H. 21 (NcPd-2) Picea wood post 5 1420 ± 60 (-23.20) 1290 – 1369 1 TO-8510 99-225 H. 21 (NcPd-2) Balaena mysticetus rib 5 1990 ± 100 CAMS-66368 99-225 H. 19 (NcPd-2) Ursus maritimus humerus 4 1510 ± 30 (-14.6) TO-8509 99-225 H. 17 (NcPd-2) Balaena mysticetus rib 4 1540 ± 70 GSC-6429 99-225 H. 17 (NcPd-2) Picea wood post 4 1210 ± 60 (-25.35) 1059 – 1185 0.874 TO-8542 99-225 H. 12 (NcPd-2) Canis familiaris tooth 4 1350 ± 60

during periods of low population levels, micro-bands 4. Our radiocarbon dates indicate a terminal Dorset occu- appear to have been the largest social units, presumably pation on western Victoria Island extending to about 600 as a response to fewer resources. BP, thus overlapping by at least a century with Thule occupations in the area. This suggestion is consistent 2. As in other areas of the Canadian Arctic and Greenland, with that of Friesen (2007) for sites on southeastern Vic- the longhouse – hearth row set complex appeared during toria Island and with the youngest Dorset dates from the Dorset period. However, it apparently appeared ear- King William Island, Boothia Peninsula, and Somerset lier on western Victoria Island than in other regions from Island (Damkjar, 2000; Dyke and Savelle, 2009; Dyke et which dates are reported. This pattern indicates that al., 2011). there may have been significant influences on the development of this complex farther west than previously rec- ognized. Alternatively, early Dorset communal features ACKNOWLEDGEMENTS on western Boothia Peninsula and at Alarnerk in Foxe Basin may represent the precursors of the more familiar Field work in 1999 – 2001 was supported by the Geologi- Late Dorset longhouses. cal Survey of Canada (GSC), the Climate Change Action Fund (Natural Resources Canada), the Polar Continental Shelf Pro- 3. The number of hearths within hearth-row sets and com- gram (Natural Resources Canada), and the Social Sciences and partments within longhouses is consistent at each site Humanities Research Council of Canada. Donald Johnson, now at which multiple hearth-row sets or longhouses occur. at Lakehead University, and Tony Alanak and Aaron Kimik- This suggests a fundamentally different, more highly sana of Holman assisted Peter Whitridge in excavating at Innirit structured, social system in operation than was operat- Point. Roger McNeely (GSC), Tim Jull (University of Arizona), ing in earlier Paleoeskimo societies. Furthermore, the Rolf Beukens (University of Toronto), and John Southon (then at pairings of large and small hearths within hearth-row the Lawrence Livermore National Laboratory, California) meas- sets suggest a heightened degree of symbolism at the ured the radiocarbon ages, and Bob Mott (GSC) identified wood communal feasting level, one not seen in Pre-Dorset or and charcoal samples. Stephen Wolfe reviewed the manuscript early Dorset societies. for GSC and three anonymous reviewers for the journal. All pro- vided comments that led to corrections and better clarity. 180 • J.M. SAVELLE et al.

REFERENCES Jenness, D. 1922. The life of the Copper Eskimos. Report of the Canadian Arctic Expedition, 1913 – 18, Vol. 12. Ottawa: F.A. Appelt, M., and Gulløv, H.C. 1999. Late Dorset in High Arctic Acland (King’s Printer). Greenland: Final report on the Gateway to Greenland Project. Kelly, R.L. 1995. The foraging spectrum: Diversity in hunter- Danish Polar Center Publication 7. Copenhagen: Danish gatherer lifeways. Washington, D.C.: Smithsonian Institution National Museum and Danish Polar Center. Press. Binford, L.R. 1980. Willow smoke and dogs’ tails: Hunter- Lyman, R.L. 1994. Vertebrate taphonomy. Cambridge: Cambridge gatherer settlement systems and archaeological site formation. University Press. American Antiquity 45:4 – 20. Maxwell, M.S. 1985. Prehistory of the eastern Arctic. Orlando: ———. 1982. The archaeology of place. Journal of Anthropological Academic Press. Archaeology 1:5 – 31. McGhee, R. 1971. An archaeological survey of western Victoria ———. 2001. Constructing frames of reference: An analytical Island, N.W.T., Canada. Bulletin 232. Ottawa: National method for archaeological theory building using ethnographic Museum of Canada. 158 – 191. and environmental data sets. Los Angeles: University of ———. 1976. Paleoeskimo occupations of central and High Arctic California Press. Canada. In: Maxwell, M.S., ed. Eastern Arctic prehistory: Birket-Smith, K. 1929. The Caribou Eskimos. Report of the Fifth Paleoeskimo problems. Memoirs of the Society for American Thule Expedition 1921 – 24, Vol. 5. Copenhagen: Gyldendal. Archaeology 31. 15 – 39. Damkjar, E. 2000. A survey of Late Dorset longhouses. In: ———. 1979. The Paleoeskimo occupations at Port Refuge, High Appelt, M., Berglund, J., and Gullov, H.C., eds. Identities and Arctic Canada. Mercury Series, Archaeological Survey of cultural contacts in the Arctic. Danish Polar Center Publication Canada Paper 92. Ottawa: National Museum of Man. 8. Copenhagen: Danish National Museum and Danish Polar ———. 1981. The Dorset occupations in the vicinity of Port Center. 170 – 180. Refuge, High Arctic Canada. Mercury Series, Archaeological ———. 2005. Late Dorset longhouses: A look inside. In: Survey of Canada Paper 105. Ottawa: National Museum of Sutherland, P.D., ed. Contributions to the study of the Dorset Man. Palaeo-Eskimos. Mercury Series, Archaeology Paper 167. ———. 1996. Ancient people of the Arctic. Vancouver: UBC Gatineau: Canadian Museum of Civilization. 147 – 165. Press. Dyke, A.S., and Savelle, J.M. 2000. Holocene driftwood incursion ———. 2000. Radiocarbon dating and the timing of the Thule to southwestern Victoria Island, Canadian Arctic Archipelago, migration. In: Appelt, M., Berglund, J., and Gullov, H.C., eds. and its significance to paleoceanography and archaeology. Identities and cultural contacts in the Arctic. Danish Polar Quaternary Research 54:113 – 120. Center Publication 8. Copenhagen: Danish National Museum ———. 2009. Paleoeskimo demography and sea-level history, and Danish Polar Center. 181 – 191. Kent Peninsula and King William Island, central Northwest Meldgaard, J. 1960. Origin and evolution of Eskimo culture in Passage, Arctic Canada. Arctic 62:371 – 392. the eastern Arctic. Canadian Geographical Journal 60:64 – 75. Dyke, A.S., Savelle, J.M., and Johnson, D.S. 2011. Paleoeskimo ———. 1962. On the formative period of the Dorset culture. In: demography and Holocene sea-level history, Gulf of Boothia, Campbell, J.M., ed. Prehistoric cultural relations between the Arctic Canada. Arctic 64:151 – 168. Arctic and temperate zones of North America. Technical Paper Friesen, T.M. 2000. The role of social factors in Dorset-Thule 11. Montreal: Arctic Institute of North America. 92 – 95. interaction. In: Appelt, M., Berglund, J., and Gullov, H.C., Morrison, D. 2000. The arrival of the Inuit: Amundsen Gulf and eds. Identities and cultural contacts in the Arctic. Danish Polar the Thule migration. In: Appelt, M., Berglund, J., and Gullov, Center Publication 8. Copenhagen: Danish National Museum H.C., eds. Identities and cultural contacts in the Arctic. Danish and Danish Polar Center. 206 – 220. Polar Center Publication 8. Copenhagen: Danish National ———. 2002. Hunter-gatherer aggregations writ large: Dorset Museum and Danish Polar Center. 221 – 228. longhouses on the Ekalluk River, Arctic Canada. Paper Murray, M.S. 1999. Local heroes. The long-term effects of short- presented at the 67th Annual Meeting of the Society for term prosperity – an example from the Canadian Arctic. American Archaeology, 20 – 24 March 2002, Denver, Colorado. World Archaeology 30:466 – 483. ———. 2004. Contemporaneity of Dorset and Thule cultures Outridge, P.M., Hobson, K.A., and Savelle, J. 2009. Long-term in the North American Arctic: New radiocarbon dates from changes of mercury levels in ringed seal (Phoca hispida) from Victoria Island, Nunavut. Current Anthropology 45:685 – 691. Amundsen Gulf, and beluga (Delphinapterus leucas) from the ———. 2007. Hearth rows, hierarchies, and Arctic hunter- Beaufort Sea, western Canadian Arctic. Science of the Total gatherers: The construction of equality in the Late Dorset Environment 407(23):6044 – 6051. period. World Archaeology 39:194 – 214. Park, R.W. 2002. Archaeological excavations at Brooman Point Hayes, M.G., Coltrain, J.B., and O’Rourke, D.H. 2005. Molecular (QiLd-1) and Hornby Head (RbJq-1), summer 2001. Unpubl. archaeology of the Dorset, Thule, and Sadlermiut: Ancestor- report on file, Nunavut Ministry of Culture, Language, Elders descendant relationships in eastern North American Arctic and Youth, PO Box 310, Igloolik, Nunavut X0A 0H0. prehistory. In: Sutherland, P.D., ed. Contributions to the study ———. 2003. The Dorset culture longhouse at Brooman Point, of the Dorset Palaeo-Eskimos. Mercury Series, Archaeology Nunavut. Études Inuit Studies 27:239 – 253. Paper 167. Gatineau: Canadian Museum of Civilization. 11 – 32. PALEOESKIMO DEMOGRAPHY ON VICTORIA ISLAND • 181

Plumet, P. 1982. Les maisons longues dorsétiennes de l’Ungava. Schledermann, P. 1978. Prehistoric demographic trends in the Géographie physique et Quaternaire 36:253 – 289. Canadian High Arctic. Canadian Journal of Archaeology Rasmussen, K. 1931. The Netsilik Eskimos: Social life and spiritual 2:43 – 58. culture. Report of the Fifth Thule Expedition, 1921 – 24, Vol. 9. ———. 1990. Crossroads to Greenland: 3000 years of prehistory Copenhagen: Nordisk Forlag. in the Eastern High Arctic. Calgary: The Arctic Institute of Rowley, S. 1994. The Sadlermiut: Mysterious or misunderstood. North America, University of Calgary. In: Morrison, D., and Pilon, J.-L., eds. Threads of Arctic ———. 1996. Voices in stone: A personal journey into the prehistory: Papers in honour of William E. Taylor, Jr. Arctic past. Calgary: The Arctic Institute of North America, Archaeological Survey of Canada Paper 149. Ottawa: Canadian University of Calgary. Museum of Civilization Mercury Series. 361 – 384. Taylor, W.E. 1972. An archaeological survey between Cape Parry Savelle, J.M., and Dyke, A.S. 2002. Variability in Palaeoeskimo and Cambridge Bay, N.W.T. Mercury Series, Paper 1. Ottawa: occupation on south-western Victoria Island, Arctic Canada: Archaeological Survey of Canada. Causes and consequences. World Archaeology 33:508 – 522. Whitridge, P., Savelle, J.M., and Dyke, A.S. 2001. Technical ———. 2009. Paleoeskimo demography on western Boothia mediation and the longhouse: Latourian translation networks Peninsula, Arctic Canada. Journal of Field Archaeology in Late Dorset prehistory. Paper presented at the 66th Annual 34:267 – 283. Meeting of the Society for American Archaeology, 18 – 22 April 2001, New Orleans, Louisiana.